Recently the microbenchmarks were flakey, but from an older bug. Turns out, `LiveTestWidgetsFlutterBindingFramePolicy` is defaulted to `fadePointers` with this fun note:
> This can result in additional frames being pumped beyond those that
the test itself requests, which can cause differences in behavior
Both `text_intrinsic_bench` and `build_bench` use a similar pattern:
* Load stocks app
* Open the menu
* Switch to `benchmark` frame policy
What happens, rarely, is that
`LiveTestWidgetsFlutterBinding.pumpBenchmark()` will call (async) `handleBeginFrame` and `handleDrawFrame`. `handleDrawFrame` juggles a tri-state boolean (null, false, true). This boolean is only reset to `null` when handleDrawFrame is called back to back, say, from an extra frame that was scheduled.
1. Switch tri-state boolean to an enum, its easier to read
2. remove asserts that compile away in benchmarks (`--profile`)
3. use `Error.throwWithStackTrace` to keep stack traces.
I've been running this test on device lab hardware for hundreds of runs and have not hit a failure yet.
Fixes#150542Fixes#150543 - throw stack!
**Three things**
Re-lands #154374
New: fix `platform_channels_benchmarks` to print the "done" key. Updated notes for the microbenchmark parser. There are no other users of `microbenchmarks.readJsonResults`.
Re-Re-land: Uninstall microbenchmarks before running them.
Flakes in https://github.com/flutter/flutter/issues/153828 stem from adb saying the app isn't installed, but then failing to install wtih -r. Several other tests uninstall the app before trying to run it. Previous fix called uninstall between tests, but iOS takes 12 to 13 seconds to perform uninstall / install, which timed out the test. Just uninstall the one time since we only care about any lingering apps with different keys.
Potential solution https://github.com/flutter/flutter/issues/153828
Re-land Make things go fast
Instead of installing 21 different compilations of the same app to get results; compile and run them together. Locally on Mac+iOS, this should takes ~3 minutes instead of ~15 minutes.
Reverts: flutter/flutter#154374
Initiated by: jtmcdole
Reason for reverting: A different benchmark was using the microbenchmark parser and timing out.
Original PR Author: jtmcdole
Reviewed By: {zanderso}
This change reverts the following previous change:
Two things:
**Re-land**: Uninstall microbenchmarks before running them.
Flakes in #153828 stem from adb saying the app isn't installed, but then failing to install wtih `-r`. Several other tests uninstall the app before trying to run it. Previous fix called uninstall between tests, but iOS takes 12 to 13 seconds to perform uninstall / install, which timed out the test. Just uninstall the one time since we only care about any lingering apps with different keys.
Potential solution #153828
**Make things go fast**
Instead of installing 21 different compilations of the same app to get results; compile and run them together. Locally on Mac+iOS, this should takes ~3 minutes instead of ~15 minutes.
Two things:
**Re-land**: Uninstall microbenchmarks before running them.
Flakes in #153828 stem from adb saying the app isn't installed, but then failing to install wtih `-r`. Several other tests uninstall the app before trying to run it. Previous fix called uninstall between tests, but iOS takes 12 to 13 seconds to perform uninstall / install, which timed out the test. Just uninstall the one time since we only care about any lingering apps with different keys.
Potential solution #153828
**Make things go fast**
Instead of installing 21 different compilations of the same app to get results; compile and run them together. Locally on Mac+iOS, this should takes ~3 minutes instead of ~15 minutes.
Reverts: flutter/flutter#152487
Initiated by: gmackall
Reason for reverting: I forgot that I need to override the compileSdkVersion in the AGP 8.0 [instance of this test](ef9cd32f5a/dev/devicelab/bin/tasks/android_java17_dependency_smoke_tests.dart (L19))
Original PR Author: gmackall
Reviewed By: {reidbaker}
This change reverts the following previous change:
Updates `compileSdk`, `targetSdk`, and `ndk` versions (former 2 to latest, latter to the version of the ndk we are hosting on CIPD).
Summary of changes:
- Updates mentioned template values
- `compileSdk` 35 requires AGP 8.0+, so updated to 8.1 in many places.
- This also necessitated Gradle upgrades in most places
- This also necessitated moving the `package` xml attribute to the AGP `namespace` field in a couple places (test + template).
- Some tests use the output of `flutter create` but then use intentionally lower AGP versions. [I downgraded the `compileSdk` in these tests.](fee34fd61a)
- [Stopped lockfile generation](82324a2570) script from hitting the `hello_world` example because it uses `.kts` gradle files.
- One test needed [some Gradle options we had already added to templates](6aa187b4b6).
Updates `compileSdk`, `targetSdk`, and `ndk` versions (former 2 to latest, latter to the version of the ndk we are hosting on CIPD).
Summary of changes:
- Updates mentioned template values
- `compileSdk` 35 requires AGP 8.0+, so updated to 8.1 in many places.
- This also necessitated Gradle upgrades in most places
- This also necessitated moving the `package` xml attribute to the AGP `namespace` field in a couple places (test + template).
- Some tests use the output of `flutter create` but then use intentionally lower AGP versions. [I downgraded the `compileSdk` in these tests.](fee34fd61a)
- [Stopped lockfile generation](82324a2570) script from hitting the `hello_world` example because it uses `.kts` gradle files.
- One test needed [some Gradle options we had already added to templates](6aa187b4b6).
Removes these two discontinued plugins from `dev/integration_tests/flutter_gallery`
[`device_info`](https://pub.dev/packages/device_info):
Apparently the video playback doesn't work on iOS simulators (I wasn't able to verify this, as I don't have an iOS simulator installed). I removed the guard against running on those simulators, and replaced with a note in the README.
[`connectivity`](https://pub.dev/packages/connectivity):
This plugin was used to play the bee video from the network. I changed the demo so that the bee video is instead also played from an asset (like its friend the butterfly), and then removed the use of the plugin.
Unblocks the re-land of https://github.com/flutter/engine/pull/53462 (itself a reland ð), because of https://github.com/flutter/flutter/pull/150465#issuecomment-2181403712.
Manual recreation of https://github.com/flutter/flutter/pull/148911
Entire PR is just the output of
```
flutter update-packages --force-upgrade
```
followed by (run from the root of the flutter repo)
```
find . -type d -name 'android' | dart dev/tools/bin/generate_gradle_lockfiles.dart --no-gradle-generation --no-exclusion
```
This benchmark illustrates the impact that const'ness has on widget build times by building a widget tree with all const-widgets and then building the same widget tree without any const'ness. Local testing with this benchmark running on a Moto G4 shows that const'ness is 13-16% faster.
I'd like to check this benchmark in because the question of how important widget const'ness is comes up every so often. With this benchmark we have up-to-date data to point people to.
DDS was temporarily pinned to 4.1.0 because 4.2.0 triggered some test
failures (see https://github.com/flutter/flutter/pull/147250). Those
failures should be fixed by vm_service 14.2.2, so this unpins DDS and
rolls both of these packages (along with devtools_shared, which is a DDS
dependency).
(If the bot updates vm_service before this is done, I can rebase over
that will reduce the size of this PR to just a few files)
## Description
This introduces a list of packages that we will explicitly not pin. It is to be used for things where the package isn't actually published, but is a transitive dependency of another package included in the SDK. This happens with the `macros` package, for instance, which depends on the private, unpublished, `_macros` package where the SDK does some tricky things to depend on it (it depends on "any", but ships it as part of the SDK).
Also ran `flutter update-packages --force-update` to update all of the pubspec files.
## Related Issues
- Fixes#147656
## Tests
- Added a test that makes sure that explicitly unpinned packages don't show up in the pinned list.
- added language for all code blocks
- replaced `bash` or `shell` with `sh` for consistency.
- added `sh` and `console` in the GitHub template link generator.
- updated test for GitHub template.
The interface for `ArgResults` from `package:args` has added new fields. Change the implementations of these in the conductor to extend `Fake` so that these aren't analyzer errors.
This broke the pub roll here: https://github.com/flutter/flutter/pull/146642#issuecomment-2050169629
This PR is to update material_color_utilities package version to the latest. `material_color_utilities/scheme/scheme_fruit_salad.dart` and `material_color_utilities/scheme/scheme_rainbow.dart` are exported after version 0.9.0.
Once this PR is merged, we don't need to explicitly import these two files like the change in PR #144805, which breaks some dependencies in `Google testing`.
Add the missing `Directionality` widget and `await test.pump()` call: commit: 0fd7712fa7
Without the `pump` 1s, it sometimes schedules multiple `handleBeginFrame`s and `handleDrawFrame`s.
For https://github.com/flutter/flutter/pull/144577. There's no promise that the performance will be great when `IntrinsicHeight/IntrinsicWidth` is used extensively but it's not that uncommon of a widget.
